Yes, I was confused. But now I think I understand.
Btw: your english is not worse than mine..
Here we go:
Signed-off-by: Daniel Alder <hac...@wolke7.net>
Index: aufs/fs/aufs/super.c
===================================================================
--- aufs.orig/fs/aufs/super.c 2012-03-20 11:00:24.660564803 +0100
+++ aufs/fs/aufs/super.c 2012-03-20 11:00:24.672564803 +0100
@@ -310,107 +310,75 @@
return ULLONG_MAX;
}
-static u64 au_add_muldiv_till_max(u64 a, u64 b, u64 mul, u64 div)
+static u64 au_mul_till_max(u64 a, long mul)
{
u64 old;
- b *= mul;
-
old = a;
- a += div64_u64(b, div);
+ a *= mul;
if (old <= a)
return a;
return ULLONG_MAX;
}
-static u64 au_div_rup(u64 a, u64 b)
-{
- u64 ret;
-
- ret = div64_u64(a, b);
- if (ret * b != a)
- ret++;
- return ret;
-}
-
static int au_statfs_sum(struct super_block *sb, struct kstatfs *buf)
{
int err;
- long biggest;
+ long bsize, factor;
u64 blocks, bfree, bavail, files, ffree;
aufs_bindex_t bend, bindex, i;
unsigned char shared;
struct path h_path;
struct super_block *h_sb;
- struct {
- long bsize;
- u64 blocks, bfree, bavail;
- } *bs;
- err = -ENOMEM;
bend = au_sbend(sb);
- /*
- * if you really need so many branches, then replace kmalloc/kfree by
- * vmalloc/vfree.
- */
- BUILD_BUG_ON(sizeof(*bs) * AUFS_BRANCH_MAX > KMALLOC_MAX_SIZE);
- bs = kmalloc(sizeof(*bs) * (bend + 1), GFP_NOFS);
- if (unlikely(!bs))
- goto out;
- biggest = 0;
+ bsize = LONG_MAX;
files = 0;
ffree = 0;
+ blocks = 0;
+ bfree = 0;
+ bavail = 0;
for (bindex = 0; bindex <= bend; bindex++) {
h_path.mnt = au_sbr_mnt(sb, bindex);
h_sb = h_path.mnt->mnt_sb;
shared = 0;
for (i = 0; !shared && i < bindex; i++)
shared = (au_sbr_sb(sb, i) == h_sb);
- if (shared) {
- bs[bindex].bsize = 0;
+ if (shared)
continue;
- }
/* sb->s_root for NFS is unreliable */
h_path.dentry = h_path.mnt->mnt_root;
err = vfs_statfs(&h_path, buf);
if (unlikely(err))
- goto out_bs;
+ goto out;
+
+ if (bsize > buf->f_bsize) {
+ /*
+ * we will reduce bsize, so we have to expand blocks
+ * etc. to match them again
+ */
+ factor = (bsize / buf->f_bsize);
+ blocks = au_mul_till_max(blocks, factor);
+ bfree = au_mul_till_max(bfree, factor);
+ bavail = au_mul_till_max(bavail, factor);
+ bsize = buf->f_bsize;
+ }
- if (buf->f_bsize > biggest)
- biggest = buf->f_bsize;
- bs[bindex].bsize = buf->f_bsize;
- bs[bindex].blocks = buf->f_blocks;
- bs[bindex].bfree = buf->f_bfree;
- bs[bindex].bavail = buf->f_bavail;
+ factor = (buf->f_bsize / bsize);
+ blocks = au_add_till_max(blocks,
+ au_mul_till_max(buf->f_blocks, factor));
+ bfree = au_add_till_max(bfree,
+ au_mul_till_max(buf->f_bfree, factor));
+ bavail = au_add_till_max(bavail,
+ au_mul_till_max(buf->f_bavail, factor));
files = au_add_till_max(files, buf->f_files);
ffree = au_add_till_max(ffree, buf->f_ffree);
}
- blocks = 0;
- bfree = 0;
- bavail = 0;
- for (bindex = 0; bindex <= bend; bindex++) {
- if (!bs[bindex].bsize)
- continue;
- if (bs[bindex].bsize != biggest) {
- bs[bindex].bsize = biggest / bs[bindex].bsize;
- bs[bindex].blocks = au_div_rup(bs[bindex].blocks,
- bs[bindex].bsize);
- bs[bindex].bfree = au_div_rup(bs[bindex].bfree,
- bs[bindex].bsize);
- bs[bindex].bavail = au_div_rup(bs[bindex].bavail,
- bs[bindex].bsize);
- }
-
- blocks = au_add_till_max(blocks, bs[bindex].blocks);
- bfree = au_add_till_max(bfree, bs[bindex].bfree);
- bavail = au_add_till_max(bavail, bs[bindex].bavail);
- }
-
- buf->f_bsize = biggest;
+ buf->f_bsize = bsize;
buf->f_blocks = blocks;
buf->f_bfree = bfree;
buf->f_bavail = bavail;
@@ -418,8 +386,6 @@
buf->f_ffree = ffree;
buf->f_frsize = 0;
-out_bs:
- kfree(bs);
out:
return err;
}
On 03/20/2012 02:14 AM, sf...@users.sourceforge.net wrote:
> Daniel Alder:
>> aufs: au_statfs_sum calculates wrong sums when different block sizes are in
>> use
>>
>> If aufs is mounted in sum mode, but the source file systems have different
>> block sizes, the result of statfs is the sum of all block counts which does
>> not make sense. This patch corrects this problem by normalizing the block
>> counts to the same block size and then adds them.
>
> That is not what I want.
> If my poor English made you misunderstood, I am sorry.
>
> What I want now is only last one patch. In my local tree, I have already
> applied your first 3 patches and 1 patch of mine. And then you posted
> another patch to modify my patch. This is the patch I want and I asked
> to add some minor changes.
> You don't have to merge all of your patches.
>
> After applying the patch, I will revert your 2/3 patch.
>
>
> J. R. Okajima
>
------------------------------------------------------------------------------
This SF email is sponsosred by:
Try Windows Azure free for 90 days Click Here
http://p.sf.net/sfu/sfd2d-msazure
